Nonmagnetic framboid and associated iron nanoparticles with a space-weathered feature from asteroid Ryugu – Nature Communications
Typical framboid
In the stored sample, the main A0064 particle (~3mm in size, mass 6.7mg) was surrounded by many tiny particles that had likely detached from the surface of this particle [Supplementary Information (SI), Fig. S1A]. Two of these tiny particles, named FO007 (17090m2) and FO008 (180100m2), were mounted onto an indium plate (SI, Fig. S1B) for easier handling and for scanning electron microscopy (SEM) observation (SI, Fig. S1C, D). The tiny particles divided into many pieces when pressed, and each piece was assigned a branch number (SI, Fig. S1E, SIF). The fact that they were easily broken is consistent with the report that Ryugu particles are brittle8. Compositional analysis by SEMenergy-dispersive X-ray spectrometry (EDS) was performed on particle A0064-FO007-I (7254m2), which had a relatively high iron content on its surface and contained many spherical subparticles resembling framboidal magnetites (Fig.1). Ultrathin sections of particle A0064-FO007-I with a thickness of 150nm (Fig.1A) were prepared by focused-ion-beam (FIB) machining for subsequent analysis (Methods).
A Secondary electron image. B Corresponding backscatter electron image. C Scanning electron microscopyenergy-dispersive X-ray spectrometry (EDS) elemental mapping at an acceleration voltage of 5kV. Note that, because the detector was located to the right of the image, the left-hand side of the particle was shaded and could not be detected accurately. D EDS corresponds to the whole area of A0064-FO007-I in region (C). The scale bars for (AC) are 10m.
The ultrathin section from position i (FIBi) contained approximately a dozen framboidal magnetite particles with sizes in the range 500900nm (Fig.2A). The magnetic domain structure of these particles was examined by electron holography, and the magnetic domains in each of the particles were found to have concentric circular magnetic structures, typical of submicrometer magnetite particles (Fig.2B)23,24,25,26.
A, B. Bright-field transmission electron microscopy (TEM) and the corresponding magnetic-flux-distribution images, respectively, of framboidal magnetite particles in a thin section prepared from Position i (FIBi) in Fig.1A. C Bright-field TEM image of a thin section extracted from the Position ii (FIBii) in Fig.1A. D Magnetic-flux-distribution image of box 2d in (C). All magnetic-flux-distribution images are two times the phase-amplified reconstruction. Scale bars are 500nm for (A, B), 2m for (C) and 1m for (D).
Nonmagnetic framboid
Ultrathin sections taken from the neighboring position ii (FIBii) contained approximately ten framboidal particles of size 400800nm (Figs.2C, SI, S2A), similar to the particles observed in sections taken from position FIBi. Surprisingly, electron holography revealed an absence of magnetic-domain structures and showed a homogeneous contrast, suggesting that the particles were nonmagnetic and differed markedly from magnetite (Figs.2D, SI, S2B,C).
To investigate the composition of these nonmagnetic particles, scanning transmission electron microscopy (STEM)EDS analysis was performed. The results show that the spherical particles contained iron and oxygen as the only major components, indicating that they consisted of iron oxides (Figs.3, 4, SI, S3). However, the signal intensity for oxygen in the nonmagnetic particles compared with that in the matrix was weaker than that in typical magnetite, suggesting that the nonmagnetic particles were formed by the reduction of magnetite. The average Fe/O ratio of the five spherical particles (SI, Fig. S3) was 0.7280.035, which is slightly smaller than that of magnetite (Fe3O4) at 0.75. However, the STEMEDS results are not highly quantitative and are consistent with magnetite within the error range.
This thin section corresponds to that in Fig.1D. The original surface is at the left-hand side of the image. Note that 30-nm-thick carbon layers were deposited onto both faces of the thin section. Oxygen in the rounded particles at the surface and sodium, magnesium, silicon, and sulfur in the matrix just beneath the rounded particle have been depleted. Tiny aluminum round spots below the surface suggest melting at a high temperature. The yellow lines are guides, indicating the surface and positions 1m and 2m below the surface. The white brackets in the iron map indicate areas where iron appears to have diffused from the spherical particles into the surrounding matrix. The scale bar is 2m.
The eleven arrows indicate metals for which the elemental ratio of iron and nickel is shown in SI TableS1. The original surface is at the left-hand side of the image. Note that a 30-nm-thick layer of carbon was deposited onto each face of the thin section. The widespread distribution of iron in the matrix surrounding the pseudo-magnetite exclusively is likely a result of its release from the framboid. The scale bar is 500nm.
We therefore used electron energy-loss spectroscopy (EELS) to analyze the bonding state of iron and oxygen (Fig.5A, B). In the EELS data of magnetite and maghemite, two characteristic peaks related to oxygen are generally observed at ~530eV and ~540eV, respectively, whereas the spectrum of wstite shows a single peak at ~540eV because of its relatively weak ~530eV peak (Fig.5A). A comparison of the spectra of magnetite, maghemite, and wstite suggests that the nonmagnetic particles in Fig.2D consist of wstite because its spectrum shows only a single peak at 540eV (c1 in Fig.5A). However, the iron peak in the EELS data of wstite is shifted toward lower energy compared with that in the spectra of magnetite and maghemite, whereas the position of the iron peak in the EELS data of our nonmagnetic particles was closer to that of magnetite than to that of wstite. Because wstite is an antiferromagnetic mineral, the nonmagnetic characteristics in the electron holography were consistent with the presence of wstite. In summary, the nonmagnetic particles display characteristics of both magnetite and wstite in terms of the bonding states between iron and oxygen. We have named such particles that display the characteristics of both wstite and magnetite and that do not exhibit a magnetic-domain structure, pseudo-magnetite.
A, B. Electron energy-loss spectroscopy (EELS) data of oxygen and iron, respectively, obtained from boxes c1 in Fig.2, d2f3. The bottom three spectra are reference spectra for synthesized wstite, hematite, and magnetite (purity: 99.5%; Kojundo Chemical Lab. Co., Ltd.), respectively. C, D Magnetic-flux-distribution and high-angle annular dark-field (HAADF)scanning transmission electron microscopy (STEM) images, respectively, for box 5c in Fig.2. E Magnetic-flux-distribution image of box 5e in Fig.2. F Magnified HAADFSTEM image of the box in (E). The STEMenergy-dispersive X-ray spectrometry elemental ratios in f1, f2, and f3 were Fe/Ni = 95.3:4.7, 98.0:2.0, and O/Mg/Al/Si/Fe=52.7:13.6:0.5:17.0:16.3, respectively. G, H EELS maps of oxygen and iron, respectively, corresponding to (F). All magnetic-flux-distribution images are two times the phase-amplified reconstruction. Scale bars are 200nm for (CE) and 100nm for (FH).
On the left-hand side of each panel of the STEMEDS mapping images of particle FIBii (Figs.3, 4, SI, S3), corresponding to the surface of the A0064-FO007-I particle, not only is the signal for oxygen reduced in intensity, but the signal intensities for various other light elements are substantially reduced. Sodium and sulfur almost disappeared from the surface to the region ~2m below the surface. Magnesium and silicon also showed a compositional decline in the 12m range from the surface. Aluminum was present as numerous small spherical particles in the 12m region from the surface, suggesting that it might have melted and resolidified at one time. In addition, the elemental map of iron shows that the region in the matrix near the pseudo-magnetite particles was enriched in iron compared with the internal matrix. We believe that this iron had been released and diffused from the pseudo-magnetite.
Metallic iron particles
Near the pseudo-magnetite particles in the same ultrathin section, vortex-like magnetic domain structures, characteristic of ferromagnetic materials, were observed (Fig.5CF). The EELS data of these particles showed an iron signal; however, the peak position was shifted to a lower energy compared with those of magnetite and wstite (Fig.5B). By contrast, the oxygen signal was very weak or almost absent (Fig.5A). The EELS maps of oxygen and iron (Fig.5G, H) show that metallic iron particles were present. The combination of these results with those of STEMEDS elemental mapping (Figs.3, 4, SI, S3) suggest that the vortex particle in Fig.5C consisted of metallic iron, whereas that observed in Fig.5E consisted of an ironnickel alloy containing 25 at% Ni. The magnetic vortex structures of the several-hundred-nanometer-sized iron particles are consistent with those in dusty olivines from the Semarkona meteorite27. The presence of many tiny iron particles without nickel near the surface of the region in Fig.5C suggests that the tiny nickel-free iron particles formed as a result of the reduction of magnetite by micrometeoroid bombardment (Fig.4, SI, TableS1). Since the iron particles in the interior contained 812 at% Ni, iron particles containing ~4 at% Ni in the intermediate region may be a result of reaction with the iron-nickel particles in the interior. Line profiles of iron and nickel from the surface to the interior are also consistent with this formation scenario involving impact heating (SI, Fig. S5).
More than one hundred metallic iron particles with a size of 30400nm were easily counted in the alteration region (shown by white brackets in Fig.2 and arrows in SI, Fig. S5) 2m in depth from the surface, 1020m in length, and 0.1m in thickness. Assuming that a single impact event alters a region 10m in diameter to a depth of 2m, the total number of iron particles produced would be on the order of ~104, which is a sufficient number to acquire remanent magnetization during an event involving the formation of iron nanoparticles by micrometeoroid bombardment. The volume of the alteration region is estimated to be (1.66.3)1016m3 by assuming a cylindrical shape, and the number density of the iron nanoparticles is the order of ~1019m3. The quantitative remanent intensity will be discussed briefly in Section Future perspectives.
Other types of pseudo-magnetite
The analysis results for ultrathin sections extracted from positions iii (FIBiii) and iv (FIBiv) of particle A0064-FO007-I are shown in Fig.6. At position iii, a spherical particle ~2.5m in diameter and consisting of two distinctly different regions of contrast was isolated above the surface of the particle (Fig.6A). Many microcrystals were found in the lower-contrast region at the top of the particle. The selected-area electron-diffraction (SAED) pattern of this region (circle b in Fig.6A) shows that it was composed of polycrystalline wstite and an amorphous material (Fig.6B). STEMEDS analysis indicates that this amorphous material was iron silicate (SI, Fig. S6). Intense spots in the SAED pattern corresponding to the particles interior (Fig.6C) indicate a single crystal of wstite; weak diffraction spots attributed to magnetite were observed at the same time (Fig.6C). Again, this particle was a pseudo-magnetite that exhibited features of both magnetite and wstite. EELS was used to examine the bonding state between oxygen and iron (Fig.6D). The oxygen peak at ~530eV hardly appeared in the spectrum corresponding to the interior of the particle or in that corresponding to the pseudo-magnetite in Fig.2. However, the peak position of iron was the same as that of wstite and differed from that of the pseudo-magnetite in Fig.2. Electron holography observation showed no magnetic domain structure characteristic of magnetite as well as the pseudo-magnetite in Fig.2 (Fig.6E). On the other hand, elemental mapping by STEMEDS showed no evidence of light-element depletion or iron leaching (Figs.6F, SI, S6). These results are not inconsistent with a scenario in which the particle was formed as a splash of a high-temperature melt heated by a micrometeoroid impact and then solidified on the surface of the particle. The surface of Ryugu, which had been exposed to interplanetary space, is enriched in silicon compared with the interior of the asteroid28. The analysis of the ultrathin section FIBiii shows the same trend (SI, Fig. S7), indicating that the surface of this particle had also been exposed to interplanetary space.
A, G High-angle annular dark-field (HAADF)scanning transmission electron microscopy (STEM) images of the thin sections extracted from Positions iii (FIBiii) and iv (FIBiv), respectively, of the surface of particle A0064FO007I, as shown in Fig.1A. B, C, I Selected-area electron-diffraction patterns from b, c in (A), and i in (G), respectively. The indices in (B, C, I) correspond to FeO (JCPDS #6-0615). The broad ring in (B) originated from amorphous iron silicate. Weak spots in (C), indicated by white arrows, correspond to magnetite. D Electron energy-loss spectroscopy data of oxygen and iron obtained from the boxes iiv in (A). E, H Magnetic-flux-distribution images corresponding to (A) and the square region in (G), respectively. F, J HAADFSTEM images and elemental mapping corresponding to (A, G), respectively. Although the white arrow in (F) indicates an intense signal for sodium, there was no characteristic peak in the energy-dispersive X-ray spectrometry data, suggesting an artificial contrast caused by the baseline. The entire data are given in SI, Figs. S6 and S8. The scale bars are 500nm for (A, E, I) and 1m for (F, G, J).
In section FIBiv, we found a nonmagnetic iron oxide (Fig.6G) that had no magnetic domain structure characteristic of magnetite (Fig.6H). Its SAED pattern corresponded to that of wstite taken from the <001> crystal zone axis (Fig.6I). The surrounding matrix did not exhibit light-element depletion (Figs.6J, SI, S8). The EELS data of oxygen and iron were consistent with that of wstite; however, STEMEDS analysis showed a higher Fe/O ratio (1.450.05) than that of wstite, which might suggest further reduction from pseudo-magnetite (SI, Fig. S9). The external shape of this particle was not spherical; however, its interior contained voids surrounded by smooth surface features that appeared to have experienced a molten state. Because the light elements were not depleted and an iron-abundant region was not present around the pseudo-magnetite, we speculate that this wstite might be the result of scattered magnetite entering the cavity of the matrix and solidifying (Figs.6J, SI, S8).
Possible space-weathering event
We surmised that pseudo-magnetite was created by reduction of magnetite as a result of space weathering by micrometeoroid impact. To confirm this hypothesis, we examined another sample, A0067, that had undergone space weathering over its entire millimeter-sized wide surface2. Because this sample had already been exposed to a strong magnetic field during conventional SEM observations in a previous study, the SEM mode associated with the FIB and conventional transmission electron microscopy (TEM) were used to search for framboids present on its surface (SI; Fig. S10AE). Although sample A0067 displayed characteristic signatures of exposure to solar wind on the asteroids surface2, we did not observe a similar texture or pseudo-magnetite in its SAED patterns (SI, Fig. S10FK), EELS data (SI, Fig. S11), or STEMEDS elemental mapping (SI, Figs. S12, S13) of framboidal particles. In addition, heating an Orgueil meteorite sample to 773K under the oxygen fugacity of the ironwstite buffer did not reduce the magnetite (SI, Fig. S14).
Magnetite coexists with wstite and iron when reduced by hydrogen at 823943K29. In addition, the texture containing melted aluminum, as shown in Fig.3, suggests that the sample had experienced heating to above the melting point of aluminum (933K). Note that the aluminum signal is not inversely correlated with oxygen and, therefore, the texture may be alumina, which has a higher melting point than aluminum. On the other hand, pyrrhotite has been observed to remain stable at temperatures as high as 543K in heating experiments30. The fact that iron sulfide and light elements such as sodium remained at depths greater than 3m from the surface of the sample (Fig.3) suggests that an event occurred that heated the top 2m of the surface of asteroid Ryugu to a temperature of ~1000K. A previous laser-irradiation experiment using an Nd:YAG laser with a wavelength of 1064nm and a pulse duration of 68ns showed that Fe nanoparticles were produced from olivine grains on a very short timescale31. Short-duration laser irradiation can simulate high-speed collisions of micrometeoroids32. The spatiotemporal scale of this experiment is similar to that of the expected impact event in the present study. The pseudo-magnetites observed in the present study were not likely produced during the sampling operation by Hayabusa2 for the following reasons. The upper limit of the peak pressure experienced by Ryugu particles during sampling operations of Hayabusa2, in which a tantalum projectile was accelerated to 30030ms1 in the sampler horn33, was estimated to be only 1.3GPa34. These events were not sufficient to heat the Ryugu particles to 300 K8. Note that the Ryugu samples analyzed in this study were collected before the artificial cratering experiment on Ryugu35. These studies support the scenario in which the pseudo-magnetite observed in the present study was produced by a micrometeoroid bombardment.
To verify the space-weathering effect of micrometeoroid bombardment, we calculated the residual temperature based on shock-physics modeling using four different parameter sets for the materials of carbonaceous asteroids8. The snapshots (right half) shown in Fig.7A suggest that the thickness of the high-temperature region is approximately equal to the diameter of the projectile, even at a high impact velocity of 50kms1, and that the temperature rapidly decreases from the surface to the interior, reaching ~600K at a thickness equal to approximately twice the diameter of the projectile at this time. The estimated range of projectile size and impact velocity was obtained by comparing the volume heated above 1000K as a function of the impact velocity (Fig.7B) to the measured volume of the pseudo-magnetite layer (Fig.7C). The projectile diameter must be smaller than the reduced region, i.e., 1020m. With increasing impact velocity, the estimated projectile size decreases because the heated volume is constrained by the observations as the volume of the reduced region. Thus, even at an impact velocity of 50kms1, the projectile size is expected to be ~2m. For Ryugu and a micrometeoroid to collide at a high velocity of >5kms1, one of them must have an eccentric orbit.
A A provenance plot (left half) and a snapshot (right half) at the time when t=50 ts, where t is the time after the initial contact and ts is the characteristic time for projectile penetration defined by the projectile diameter divided by the impact velocity. The impact velocities are shown in each panel. The spatial scale is normalized by the projectile radius. The target is a flat surface. Residual temperatures in the region of maximum pressure <1GPa are not calculated (white area). The yellow-colored regions indicate residual temperatures greater more than 1000K. B Heated volume above 1000K normalized by the projectile one as a function of impact velocity. We used the expected residual temperatures (see Methods) to estimate the heated volumes at given impact velocities. The results depend on the choice of the EOS models. Two dashed black lines are the upper and lower bounds. C The estimated projectile diameter as a function of impact velocity. We used the two dashed lines in the (B) in this calculation. The hatched region is an allowable range of the combination between the projectile diameter and impact velocity. The heated volume on the Ryugu particle is estimated to be (1.66.3)1016m3 by assuming a cylindrical shape with the diameter of (1020)m and the thickness of 2m. The dimensions were determined with the microscopic images shown in Fig.3.
The impact velocity distribution of micrometeoroids strongly depends on their origin (asteroids or comets) and the values, where is the ratio between the force due to solar-radiation pressure and that due to gravity. At least in the current Solar System, the velocity with which micrometeoroids with a radius of ~2m (~1013kg), which corresponds to <0.2, impact asteroids located between 1.5 and 4.0 astronomical units (AUs) from the Sun can be greater than 10kms136 because of the radiation pressure. Ryugu formed in the outer part of the Solar System and may have already moved inward by the time of aqueous alteration7. Therefore, the micrometeoroid would have had a more eccentric orbit at the time of impact. The presence of a micrometeoroid with an eccentric orbit suggests not only that material from the inner Solar System diffused outward37 but also that a large amount of material from the outer Solar System diffused inward. This is consistent with the finding of Fe4N in the space weather layer of a Ryugu particle, which has been interpreted as a product of a reaction between metallic iron and ammonia that was provided by outer Solar System dust3.
Reduction sequences
There are nonmagnetic framboids that show different peak positions each other in the EELS data. A nonmagnetic framboid in box 5c of Fig.2C has a peak at 707eV (spectrum d2 in Fig.5B), and a nonmagnetic framboid in box 2d of Fig.2C has a peak at 708eV (spectrum c1 in Fig.5B) in the iron EELS data. The feature at 707eV is in good agreement with those attributed to FeO (Fig.5B) and Fe2SiO438, which are Fe2+ reference materials. On the other hand, the feature at 708eV is in good agreement with the feature attributed to Fe3O4 (Fig.5B), which indicates that Fe2+/Fe3+=1/2. Because both are nonmagnetic framboids, the Fe2+/Fe3+ ratio does not appear to correlate directly with the presence of magnetism. However, oxygen spectra c1 and d2 are very similar (Fig.5A). Both show the main peak at ~540eV attributed to iron oxides without the pre-peak at ~530eV observed in the spectra of Fe3O4 and Fe2O3. The lack of the pre-peak at 530eV is consistent with the characteristics of FeO. Comparing the EELS peak positions of iron in the d2 region in Fig.5D and the c1 region in Fig.2C reveals that the framboid in the d2 region is more reduced than that in the c1 region because the c1 spectrum more closely resembles that of magnetite, whereas the d2 spectrum more closely resembles that of wstite. That is, the nonmagnetic framboid in box 5c is closer to FeO than the nonmagnetic framboids in the box 2d region in Fig.1C. In summary, the reduction of magnetite by a micrometeorite impact first results in the disappearance of the pre-peak in the EELS data for oxygen, followed by a decrease in the amount of Fe3+ and a predominance of Fe2+. The loss of magnetism is considered to occur with the disappearance of the oxygen pre-peak in the EELS data.
Future perspectives
Magnetite and wstite have similar cubic structures. The diffraction pattern of wstite shows peaks at 0.249 (111), 0.215 (200), and 0.152 (220)nm (hkl indices in brackets), and that of magnetite is similar, with peaks at 0.242 (222), 0.210 (400), and 0.148 (440)nm. Thus, magnetite and wstite, especially as framboids, are difficult to distinguish from each other in terms of their crystal shape, composition, and conventional electron-diffraction pattern. However, a comparison of Fig.2B, D shows that the magnetic properties of the two minerals are differ substantially and can assist in their detailed analysis. Pseudo-magnetite might account for only a few percent of the total magnetite. Nevertheless, the detection of its presence will be important for precise interpretation when studying the paleomagnetism and microstructure of surface materials in meteorites and extraterrestrial samples retrieved by spacecraft such as OSIRIS-Rex, because pseudo-magnetite can serve as an indicator of specific alteration processes that the sample has undergone and its presence can be taken into account to adjust the interpretation of the magnetic history of the sample.
The iron particle precipitation associated with a micrometeoroid impact could play an important role as the remanence acquisition event for carbonaceous chondrites. Because iron particle precipitation is restricted in the thin surface layer of the impacted grain, the unaltered magnetite and pyrrhotite below the thin layer dominate the magnetic signal of grains larger than ~10m. However, the magnetic measurements for smaller grains would detect the remanence record of iron particle precipitated with the micrometeoroid impact. Moreover, the remanence component of metallic iron could be distinguished from those of magnetite and pyrrhotite by stepwise demagnetization treatments. The number of iron particles contained in the altered region is roughly estimated using the observed number density (1019m3) in altered area (10m in diameter and 2m in depth) as ~104. The 104 iron particles with a diameter of 100nm had a total volume of ~51018m3 and a total saturation magnetization of ~91012Am2. The remanence intensity of the iron particles is estimated to be ~1013Am2, assuming that the intensity ratio of remanent magnetization to saturation magnetization is similar to that of thermoremanent magnetization of fine-grained magnetite acquired in a 100T field39 and to be 0.01. Although the remanence intensity should depend on the external-field intensity and the remanence-acquisition process, the precipitation of iron particles could make the remanence detectable by a sensitive magnetometer, such as a scanning microscope with a superconducting quantum interference device40. In terms of statistical thermodynamics, the paleomagnetic samples should contain a sufficiently large numbers of ferromagnetic grains to obtain accurate paleomagnetic data41. An X-ray photoemission electron microscopy study on the Imilac and Esquel pallasite meteorites demonstrated that the regions containing ~104 grains recorded the systematic changes in a dynamo magnetic field42, suggesting that the ~104 iron grains have sufficient ability to record paleomagnetic information. However, the estimation of the statistical error of paleomagnetism assuming specific conditions showed that the analyzed regions in these meteorites were not large enough to obtain accurate paleomagnetic data41. A number of collisions affecting a wider area or multiple events occurring during the timescale such that the external field could be regard as the same condition would give rise to a strong remanence intensity, resulting in increased detectability. The magnetic-field environment of the early Solar System has been reconstructed from the remanence records acquired as magnetite and pyrrhotite formed by aqueous alteration within asteroids13,14,15,16,18,19,21, magnetite formed by igneous processes20, and iron particles formed during chondrule formation12,17,27. The magnetic-field record of iron particles at the time of a collision event in a C-type asteroid could provide a spatiotemporally distinct constraint on the magnetic-field environment such as the dynamo magnetic field of an asteroid43. Note that, if the magnetite recording the natural remanence transformed into pseudo-magnetite/wstite, the transformed magnetite particles does not contribute to the paleointensity estimation based on the remanence intensity ratio of natural remanence/laboratory remanence because the pseudo-magnetite/wstite does not carry any remanences.
Micrometeoroid bombardments at a high impact velocity (>5kms1) are expected to have commenced after the dissipation of the protoplanetary disk gas, because a high-velocity micrometeoroid impactor must have an eccentric orbit, which would cause its particles to be vaporized as a result of ablation by aerodynamic heating in the case of the presence of the disk gas. The iron nanoparticles surrounding the pseudo-magnetite therefore provide a record of the magnetic field after the dissipation of the disk gas. The dynamo magnetic field of an asteroid is one important candidate of external field in such a period. In addition, sub-microscopic magnetite and associated metallic iron nanoparticles within iron sulfide were identified in lunar soil brought back by ChangE-538. The iron particles are thought to have been produced by eutectic reactions within oxygen-dissolved iron sulfide grains after reaction of molten iron sulfide with silicate gas during a large impact. The metallic iron particles were found to be magnetic. Thus, even after the dissipation of the disk gas, dynamo magnetic fields in small bodies may be imprinted during the impact of micrometeorites on the surface of airless bodies. A more comprehensive discussion of the formation epoch of nonmagnetic framboids and associated iron nanoparticles is expected in the future.
Micrometeoroid impacts similar to the one that we discovered in the present study are thought to have occurred frequently. However, these impacts have not been well understood in previous studies based on meteorites because the traces left by these events are only found on the uppermost surfaces of asteroids. This might be because, even if a meteorite fell to Earth with pseudo-magnetite or iron particles remaining, these particles would subsequently be oxidized by terrestrial weathering, although minor iron particles have been found in some carbonaceous chondrites despite them undergoing aqueous alteration44. In addition to our acquisition of the Hayabusa2 samples, retrieving samples from the asteroid Bennu by OSIRIS-REx would give us a chance to analyze them. At that time, we should avoid the remanence modification due to the oxidation of pseudo-magnetite and metallic iron nanoparticles.